Seaside redirects, really?

Sebastian,

 

Car and airplane crashes are also a bit of an exception case, yet so much effort goes into development and testing of safety systems. I’m not sure what your end game is in this thread? If you’re trying to convince the rest of us to remove the part of the framework that’s important to a lot of people, you likely will not succeed. But if you’re trying to understand the reasons for the redirect so you can make informed decisions for your own applications, by all means.

 

-Boris

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Sebastian Sastre
Sent: Monday, October 03, 2011 7:52 AM
To: Seaside - general discussion
Subject: Re: [Seaside] Seaside redirects, really?

 

On Oct 3, 2011, at 8:29 AM, Julian Fitzell wrote:

Similarly the redirects serve a well

thought purpose.

 

which is what exactly?

can you describe that well thought purpose?

 

Decouple the callback processing from the rendering.

Thus ensuring that (a) hitting refresh doesn't re-trigger an action
and (b) you don't get the browser dialog asking if you want to
resubmit form data when you hit the Back button.

 

okay those are reasons.

 

Questions:

for a: the refresh is an exception case, not the normal case, yet everything gets penalized because of that?

for b: the browser opens that dialog only on POST requests only when the user press the back button, again an exceptional case but seaside it's penalizing every other request because of that?

 

sorry but that sounds right or "well thought" to you?

 

sebastian

 

o/

 

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Well ive never run into any problems with the redirects.

But speedups are always good.  What would you suggest to fix them?

how did you measure those numbers? Did you have keep-alive active? 

With keep-alive there is only little more that is happening when doing this redirect scenario. One of the biggest latency factors is initiating a TCP connection. Most pages load quite a few images and such in parallel (today it is 4 concurrent connections or more). From that point the textual transfer of the form and the response shouldn't matter that much. With keep-alive especially that redirect scenario should be well supported. For everything else you need HTTP pipelining which is unfortunately still not used in todays browsers. 

Norbert 

which is what exactly?

can you describe that well thought purpose?

Decouple the callback processing from the rendering.

Thus ensuring that (a) hitting refresh doesn't re-trigger an action

and (b) you don't get the browser dialog asking if you want to

resubmit form data when you hit the Back button.

Seems like this ought to be easily doable with a boolean somewhere. Did we run the callbacks for continuation asdfa123? If so, don't run them. No redirect necessary, no double callbacks being run on refresh. Still have the "do you want to resubmit this form" alert, but I've never been concerned about that and don't particularly understand why people are.

Pat

yes, I confirm: this was with the keep-alive header. I'm not sure about the initiating socket thing, this should be confirmed but I guess that just making 2 DNS lookups instead of one is bad enough to eclipse that problem.

And I still cannot tell you if
- redirects have ever helped me
- redirects have ever hurt me

Just a simple FWIW.

Cheers,
Bob

Sebastian,

 

You can’t get rid of latency of both trips, just the extra one, i.e. if your single roundtrip time is 50ms then removing redirect shortens your overall time by 50ms, not by 100ms.

 

-Boris

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Sebastian Sastre
Sent: Tuesday, October 04, 2011 12:28 PM
To: Seaside - general discussion
Subject: Re: [Seaside] Seaside redirects, really?

 

On Oct 4, 2011, at 12:13 PM, Bob Arning wrote:

And I still cannot tell you if
- redirects have ever helped me
- redirects have ever hurt me

Just a simple FWIW.

Cheers,
Bob

 

in one hand it got you here.

 

In the other:

 

what's the average ping going from:

1. where your audience is to

2. where your server is located?

 

that duration times 2 is the minimum default latency you didn't get rid of

 

sebastian

 

o/

 

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

On Oct 4, 2011, at 12:13 PM, Bob Arning wrote:

And I still cannot tell you if
- redirects have ever helped me
- redirects have ever hurt me

Just a simple FWIW.

Cheers,
Bob

in one hand it got you here.

In the other:

what's the average ping going from:

1. where your audience is to

2. where your server is located?

that duration times 2 is the minimum default latency you didn't get rid of

And still what you are saying is correct that whatever I tell you the extra request goes on top. I just want to add that the effect is not as big as you might think.

Norbert